© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 2-RES-SM
    11. 10886
    Prev
    Next
    INDEX
    Explanations

    disorganized or unconventional formatting in text

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/2-res-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.2.hook_resid_post
    Hook Layer
    2
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    410
    +0.13
    0.7%
    21
    +0.12
    0.7%
    473
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    393
    +0.13
    0.04
    271
    +0.12
    0.00
    77
    +0.11
    0.02
    Negative Logits
     worn
    -1.82
    omy
    -1.72
    ,[@
    -1.59
    jam
    -1.51
     ($
    -1.45
    arm
    -1.45
     repair
    -1.41
     (£
    -1.38
     fide
    -1.36
     performed
    -1.35
    POSITIVE LOGITS
    Īĺ
    3.61
    ¿
    3.48
    Ĭ
    3.44
    ĻĤ
    3.36
    ¬
    3.35
    ľĵ
    3.26
    ´
    3.23
    ²
    3.15
    ·¸
    3.15
    «
    3.14
    Activations Density 0.260%

    No Known Activations